Skip to content

Enable FP8 quantization in SDXL using INC #1337

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
splotnikv wants to merge 2 commits into from
Closed

Enable FP8 quantization in SDXL using INC #1337

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3bbcf34
Enable FP8 quantization in SDXL using INC
splotnikv Sep 17, 2024
ac4de48
Replace command line parameter by enc variable for SDXL quant control
splotnikv Nov 27, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions examples/stable-diffusion/text_to_image_generation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -411,6 +411,7 @@ def main():
kwargs_call["image"] = control_image

# Instantiate a Stable Diffusion pipeline class
import habana_frameworks.torch.core as htcore
if sdxl:
# SDXL pipelines
if controlnet:
Expand Down
29 changes: 28 additions & 1 deletion optimum/habana/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -479,6 +479,23 @@ def __call__(
`tuple`. When returning a tuple, the first element is a list with the generated images.
"""

import os
quant_mode="disable"
quant_config_path = os.getenv('QUANT_CONFIG')
if quant_config_path != None:
import habana_frameworks.torch.core as htcore
htcore.hpu_set_env()

from neural_compressor.torch.quantization import FP8Config, convert, prepare
config = FP8Config.from_json_file(quant_config_path)
if config.measure:
quant_mode="measure"
self.unet = prepare(self.unet, config)
elif config.quantize:
quant_mode="quantize"
self.unet = convert(self.unet, config)
Copy link
Copy Markdown

@kumarans-ai kumarans-ai Sep 20, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

All Unet steps should not be converted to Fp8, Recipe for good accuracy includes last few steps in bf16. So, you must have the bf16 Unet support - refer Model Garden reference for Fp8- HQT to INC conversion patch

Copy link
Copy Markdown
Contributor Author

@splotnikv splotnikv Nov 27, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ok, I'll submit it as a separate PR. Original intention of this PR is to show how easy it is to quantize model, so I prefer to keep code change to bare minimum.

htcore.hpu_initialize(self.unet, mark_only_scales_as_const=True)

callback = kwargs.pop("callback", None)
callback_steps = kwargs.pop("callback_steps", None)

Expand All @@ -495,7 +512,13 @@ def __call__(
"Passing `callback_steps` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
)

with torch.autocast(device_type="hpu", dtype=torch.bfloat16, enabled=self.gaudi_config.use_torch_autocast):
from contextlib import nullcontext
if quant_mode == "disable":
ctx_manager = torch.autocast(device_type="hpu", dtype=torch.bfloat16, enabled=self.gaudi_config.use_torch_autocast)
else:
ctx_manager = nullcontext()

with ctx_manager:
# 0. Default height and width to unet
height = height or self.default_sample_size * self.vae_scale_factor
width = width or self.default_sample_size * self.vae_scale_factor
Expand Down Expand Up @@ -816,6 +839,10 @@ def __call__(

hb_profiler.stop()

if quant_mode == "measure":
from neural_compressor.torch.quantization import finalize_calibration
finalize_calibration(self.unet)

speed_metrics_prefix = "generation"
speed_measures = speed_metrics(
split=speed_metrics_prefix,
Expand Down